26838 research outputs found
Sort by
Succes voor TU_e meerdere NWO Perspectief-subsidies toegekend waaronder €2 9 miljoen voor AI4Optics - TUE-8 December 2025
No full text
QoE evaluation of remote physiotherapy in volumetric video and video-based real-time communication
Full text linkIn recent years, video conferencing platforms have become powerful tools for remote communication. There has also been an increase in the use of VR systems for communication. However, very few of these systems utilize photorealistic human representation. This paper investigates the strengths, challenges, and limitations of a novel 3D communication prototype (VR2Gather) and a well-established video conferencing system (Zoom). Specifically, we explore whether the 3D communication prototype can achieve comparable performance levels in a remote physiotherapy use case. By assessing various aspects, such as audio-visual quality, presence, and interaction, we aim to determine if the current prototype is comparable with commercial systems in some dimensions while exceeding expectations in others. Our results indicated that VR2Gather has the potential for a better sense of connection and higher concentration. However, challenges like improving 3D rendering quality and communication ease still need to be overcome to make it suitable for physiotherapy
Evaluating the quality of multiple automatically produced segmentation variants of the prostate on Magnetic Resonance Imaging scans for brachytherapy
Full text linkBackground and Purpose: Recently, we introduced a novel Deep Learning (DL) based (semi-)automatic method for medical image segmentation. Unlike classical DL segmentation methods, it produces multiple segmentation variants (reflecting the variation of manual segmentations) instead of just one. Potentially, with this approach, there is a higher chance that a clinician prefers one of automatically produced segmentation variants. This work focuses on evaluating this method on prostate segmentation in MRI scans used for brachytherapy and investigating its potential clinical usefulness. Materials and Methods: Three experienced radiation oncologists graded (per-slice and per-scan) segmentations produced by our method, reference segmentations (manually created and used for brachytherapy treatment planning) and segmentations produced by a classical DL method. The study was retrospective and the way the segmentation was generated (our method, classical DL method, or manually) was blinded for the clinicians. The grades reflect the amount of manual correction required. Additionally, the clinicians were asked to rank segmentations to evaluate which one is preferred for each scan. The study was performed on 13 prostate cancer patients. Results: Segmentations produced by our method are graded as requiring no manual correction in 292/576 (51 %) slices compared to 240/576 (42 %) slices in the case of the segmentations produced by a classical DL method. Furthermore, in fewer slices, 38 (6.6 %) vs. 48 (8.3 %), segmentations by our method were graded as unacceptable. Conclusion: Our study has demonstrated that deep-learning-based segmentation methods can produce high-quality segmentations. Our method was evaluated better than a classical DL method, indicating the potential for integration into clinical practice
Reproducibility report for ACM SIGMOD 2024 Paper: “ALP: Adaptive Lossless Floating-Point Compression”
Full text linkThis report describes the reproducibility evaluation carried out for the research paper "ALP: Adaptive Lossless Floating-Point Compression", authored by Azim Afroozeh, Leonardo Kuffó, and Peter Boncz from the Centrum Wiskunde & Informatica (CWI), located in Amsterdam, Netherlands. The paper, presented at SIGMOD 2024, introduces ALP, a novel adaptive lossless compression algorithm for floating-point datasets. The results of the reproducibility evaluation confirm the core contributions of the original paper, showing that ALP consistently outperforms state-of-the-art techniques in both compression ratio and (de)compression speed
Dynamic dimensioning of frequency containment reserves: The case of the Nordic Grid
No full textOne of the main responsibilities of a Transmission System Operator (TSO) operating an electric grid is to maintain a designated frequency (e.g., 50 Hz in Europe). To achieve this, TSOs have created several products called frequency-supporting ancillary services. The Frequency Containment Reserve (FCR) is one of these ancillary service products. This article focuses on the TSO problem of determining the volume procured for FCR. Specifically, we investigate the potential benefits and impact on grid security when transitioning from a traditionally static procurement method to a dynamic strategy for FCR volume. We take the Nordic synchronous area in Europe as a case study and use a diffusion model to capture its frequency development. We introduce a controlled mean-reversal parameter to assess changes in FCR obligations, in particular for the Nordic FCR-N ancillary service product. We establish closed-form expressions for exceedance probabilities and use historical frequency data as input to calibrate the model. We show that dynamic dimensioning approaches for FCR have the potential to significantly reduce exceedance probabilities (up to 37%) while maintaining the total yearly procured FCR volume equal to that of the current static approach. Alternatively, a dynamic dimensioning approach could significantly increase security at limited extra cost
Efficient join order for constraint automata through LLM-generated heuristics
No full textReo is an exogenous coordination language designed for component-based systems based on channel-based connectors. Constraint automata is defined by Christel Baier et al. as the compositional operational semantics of Reo. Semantics of a Reo circuit is computed by joining the constraint automata of the connector elements. This computation can be costly when dealing with large connectors, making an improvement necessary. Improving this operation involves either improving the join algorithm or selecting a joining order that minimizes intermediate automata. While alternative algorithms for joining constraint automata have been proposed, identifying an efficient joining order remains a challenge. This paper proposes a heuristic-based approach for finding an efficient order of joining constraint automata. By feeding OpenAI’s ChatGPT with data on the join algorithm and the structure of constraint automata, we ask it to generate diverse heuristics to identify the most efficient joining order and employ its suggestions. Our results demonstrate the impact of join order on the operation’s performance. We analyze these results to identify the best heuristic for each set of CAs based on their characteristics. This highlights the potential of LLM-driven approaches in assisting the development of efficient solutions for computational tasks
Faster approximate elastic-degenerate string matching – Part A
Full text linkAn elastic-degenerate (ED) string is a sequence of finite sets of strings. The cardinality of is the total number of strings in , for all . The size of is the total length of all strings of . ED strings have been introduced to represent a set of closely-related DNA sequences. Let be a pattern of length and be an integer. We consider the problem of -Approximate ED String Matching (EDSM): searching -approximate occurrences of in the language of . We call -Approximate EDSM under the Hamming distance, -Mismatch EDSM; and we call -Approximate EDSM under edit distance, -Edit EDSM. Bernardini et al. (Theoretical Computer Science, 2020) showed a simple -time algorithm for -Mismatch EDSM and an -time algorithm for -Edit EDSM. We improve the dependency on in both results, obtaining an -time algorithm for -Mismatch EDSM and an -time algorithm for -Edit EDSM. Bernardini et al. (Theory of Computing Systems, 2024) presented several algorithms for 1-Approximate EDSM working in time. They have also left the possibility to generalize these solutions for as an open problem. We improve the runtime of their solution for 1-Mismatch and 1-Edit EDSM from to . We further show algorithms for -Approximate EDSM for the Hamming and edit distances working in time, for any constant . Finally, we show how our techniques can be applied to improve upon the complexity of the -Approximate ED String Intersection and -Approximate Doubly EDSM problems that were introduced very recently by Gabory et al. (Information and Computation, 2025)
The FastLanes file format
Full text linkThis paper introduces a new open-source big data le format, called FastLanes. It is designed for modern data-parallel execution (SIMD or GPU), and evolves the features of previous data formats such as Parquet, which are the foundation of data lakes, and which increasingly are used in AI pipelines. It does so by avoiding generic compression methods (e.g. Snappy) in favor of lightweight encodings, that are fully data-parallel. To enhance compression ratio, it cascades encodings using a flexible expression encoding mechanism. This mechanism also enables multi-column compression (MCC), enhancing compression by exploiting correlations between columns, a long-time weakness of columnar storage. We contribute a 2-phase algorithm to nd encodings expressions during compression. FastLanes also innovates in its API, providing flexible support for partial decompression, facilitating engines to execute queries on compressed data. FastLanes is designed for ne-grained access, at the level of small batches rather than rowgroups; in order to limit the decompression memory footprint to t CPU and GPU caches. We contribute an open-source implementation of FastLanes in portable (auto-vectorizing) C++. Our evaluation on a corpus of real-world data shows that FastLanes improves compression ratio over Parquet, while strongly accelerating decompression, making it a win-win over the state-of-the-art